Exactly Gibbo, it's what most of the 'true' enthusiasts here have been saying since release.
Now we have solid figures, if anything, as confirmed by the Vortez 680 review, Nvidia's performance dominance has been reduced to almost nothing by AMD this round.
At the end of the day, nobody can complain about the performance available with either card.
This isn't the case at all, Nvidia's inability for, not even close to the first time has let AMD potentially have the better card out this round, they haven't reduced the gap to nothing.
Last time around AMD/nvidia had a core around about 360mm2, and Nvidia had another one at 530mm2, AMD were 15% behind in performance vs a core 40% bigger, and beat the similarly sized 560ti comfortably.
This round Nvidia's just over 300mm2 card is matching AMD's just over 350mm2 card, that is a massive gain from Nvidia to AMD, not the other way around.
It's not quite that simple because, nvidia has stripped out a LOT from kepler in terms of compute and offloaded significant work to the CPU(I do wonder how much power the cpu uses via drivers in a game vs a 580gtx), this helped cut out a bunch of hardware, which helped chop out a significant portion of die space, AND aiming it at 1080p really with a 256bit bus also helped it die size wise.
Make no mistake, pitcarn is more efficient than Kepler even with hardware scheduling on die and better compute performance, AMD haven't "lost" any ground, the 7970 is just a very different beast, and not well concieved.
Nvidia went from a circa 360mm2 size core, took out compute, hardware scheduling which likely dropped the size below 300mm2, and then re-added more raw horsepower in to bring the size back up. Think of it this way a 560ti with compute and hardware scheduling off die wouldn't match 6970 performance, it would have needed another 15-20% shader power.
So you can think of a 680gtx as this gen's 560ti, with 30% of the core moved off die, and 10-15% readded as more shaders, you get a highly efficient and VERY focused gaming card, it IS great.
But still not the whole story, the 7970 kept the same die size as the 6970, moved scheduling on die, added more compute AND added a bigger bus, which really isn't used except for compute and eyefinity(where it does spank the 680gtx once you factor in 680gtx doing half frame rate on 2 out of 3 screens and still being beaten... essentially 7970 offers the same performance byt the 680gtx is effectively driving 2 screens instead of three(3 screens at 60fps each, 180fps vs 1 screen at 60fps, and 2 at 30fps, giving 120fps, in other words the 7970 is really giving 50% better performance in eyefinity ).
Nvidia did the "right" thing, gamers mostly use 1080p, and mostly game, pitcarn is what AMD should have made for "gamers" at the same size as Kepler and AMD would have won this round hands down, likewise the 7970 should have been a 400-430mm2 core and 20% faster, that's life. AMD wanted to fit in everything including the kitchen sink but not comprimise on die size.
GK110 will lucky suffer the same "problems" as AMD, hardware scheduling(quite probably due to compute), more cache maybe, bigger mem bus, its likely the massively bigger die, same way as AMD, will be mostly eaten up by not gaming specific hardware

It will CRUSH a 680gtx in tri screen gaming, and will be fast, but not as much faster as you'd hope given its size.
AMD/Nvidia really should make HPC/GPGPU only cards, at stupid sizes with stupid costs, and make gaming cards for gamers. AMD or Nvidia making a 400-450mm2 core with 384bit bus, little to no compute and software scheduling = huge bags of awesome.
EDIT:- of course with a £315 Asus 7950 Direct CU, I know which way I'm going myself
